Repository navigation
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #11178 +/- ##
===========================================
- Coverage 100.00% 96.96% -3.04%
===========================================
Files 1 4 +3
Lines 24 66 +42
Branches 0 1 +1
===========================================
+ Hits 24 64 +40
- Misses 0 1 +1
- Partials 0 1 +1
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
hua7450
left a comment
There was a problem hiding this comment.
Thanks, David. Adding back each kind of business, rental and partnership loss separately follows 32 V.S.A. 6061(18)(A). (18) opens "Notwithstanding subdivisions (4) and (5)" and, unlike (5)(A), allows no netting. The candidate entry quotes the statute, RCC-146 line 13 and the booklet accurately, and Case 7 is right. The deduction add-backs are filtered against the same federal list AGI uses, so nothing is counted twice, and the hand-checked values match. One fix before I approve.
Before merge
- Capital losses follow the form while the other losses follow the statute. The formula adds back only
limited_capital_loss, the net loss on Form 1040 line 7, as RCC-146 line 13 and the 2025 booklet do. (18)(A) lists "capital loss" without defining it. In the federal code, a capital loss is each loss on a sale or exchange (26 U.S.C. 1222(2), (4)), and "net capital loss" is a separate term (1222(10)). Read the same way as the business losses, the statute adds back every capital loss deducted in AGI, so Case 6 would be 32,000 instead of 30,000. The candidate entry says the model follows the statute for "a Form 1040 or Schedule 1 line", which isn't true for line 7. Please do one of these:- add
capital_losses_allowed_against_gainsto the add-back, so capital losses follow the statute like the rest (my preference, since it matches the PR's own reading); or - keep the net figure and extend the candidate entry to say line 7 follows the form, and why.
- add
Optional
- A dependent's tax-exempt interest (follow-up).
tax_exempt_interest_incomeinaddsstill sums dependents. RCC-146 line 12 reads the claimant's own Form 1040 line 2a, the same reasoning the PR applies to Social Security. This was already true on main and the PR body discloses it, so a follow-up is fine. If you prefer, summing it over the head and spouse here is one line. #page=1on RCC-146. It is a one-page PDF, so repo style drops the fragment.- Joint-return case. Every new case is a single filer or a filer with a dependent. One married case would pin that a spouse's loss, deductions and Social Security count while a dependent's don't.
ff0618b to
b84ccf1
Compare
|
Thanks, Ziming. Fixed in b84ccf1, rebased on main. I took your preferred option. Fixed (your "Before merge" item)
Skipped (optional, not in this round): a dependent's tax-exempt interest (follow-up), Verification:
|
hua7450
left a comment
There was a problem hiding this comment.
Thanks, David. Adding back capital_losses_allowed_against_gains next to limited_capital_loss does what you describe. With limited_business_loss, the three add-backs equal loss_ald with no overlap; a carryover above the limit stays out, and dependents are excluded. Case 6 is 32,000 by hand, and the cross-spouse case now adds back the full loss. One gap left before I approve.
Before merge
-
A loss netted against the same person's gain is still not added back. The new term catches losses netted against gains outside a person's
capital_gains: the spouse's gains, or distributions without Schedule D. It misses netting inside one person'scapital_gains, which addsshort_term_capital_gainsandlong_term_capital_gainstogether.We ran one diagnostic: a 2025 single filer with 30,000 of wages, a 4,000 capital loss and a 2,000 gain, entered four equivalent ways. Federal AGI is 28,000 each time. Case 6's layout and the joint layout give 32,000. A short-term loss against the filer's own long-term gain gives 30,000, and so do distributions entered on Schedule D line 13. The new register note says these "cannot be separated", but the holding periods are separate inputs. At the prior head all four gave 30,000, so the fix introduced the split. In Addison, each such dollar is worth about $0.047 of credit.
Please either:
- add the within-person offset for the head and spouse, so every capital loss included in AGI comes back as the entry says (my preference; Montana's
mt_elderly_homeowner_or_renter_credit_gross_household_income.pyalready splits short-term from long-term and handles the carryover); or - keep the netting and correct the comment and register note to say that losses netted inside one person's
capital_gainsfollow the RCC-146 net figure.
Either way, please add a case with a short-term loss and the same person's long-term gain.
- add the within-person offset for the head and spouse, so every capital loss included in AGI comes back as the entry says (my preference; Montana's
Optional
- PR body. It still says losses netted against gains aren't added back and that Case 6 is 30,000.
- Reforms. Unlike the deduction add-back, the loss add-back isn't limited to what federal AGI actually deducted, so a reform that removes
loss_aldwould overstate income. The baseline is unaffected. - From the first round, still open: a dependent's tax-exempt interest,
#page=1on RCC-146, and a joint-return case.
32 V.S.A. 6061(18) defines renter credit income as federal AGI increased by the business, rental, farm and capital losses AGI deducted (A) and the listed above-the-line deductions (D). Neither was added back. The 75% Social Security term now reads only the claimant's and spouse's nontaxable benefits, as 6066(b)(1) and Form RCC-146 line 11 do, not a dependent's. Moves the four Vermont renter credit cases from the #10812 audit into the variable's test file and records the line 13 netting question as a source-conflict candidate. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… in Vermont renter credit income Add capital_losses_allowed_against_gains to the vt_renter_credit_income add-backs next to limited_capital_loss, so both capital loss parts that federal AGI deducts are restored. Update test Case 6 to assert both capital terms, and update the renter-credit-loss-add-back-netting source-conflict candidate note and example to match. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…er's gains in Vermont renter credit income Vermont renter credit household income now adds back capital losses that were netted against the same person's capital gains, including the prior-year long-term loss carryover absorbed by that person's gain. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
b84ccf1 to
3d9360b
Compare
|
Thanks, Ziming. Fixed in 3d9360b, rebased on main. I took your preferred option and updated the PR body to match. Fixed (your "Before merge" item)
Skipped (optional, not in this round):
Verification:
|
hua7450
left a comment
There was a problem hiding this comment.
Thanks, David. vt_renter_credit_capital_loss_add_back now adds back the loss netted inside one filer, and the diagnostic gives 32,000 for all four layouts from last round, including the same person's short-term loss against their own long-term gain and Schedule D line 13 distributions. It doesn't overlap capital_losses_allowed_against_gains or limited_capital_loss; losses above the cap stay out, and dependents are excluded. Cases 11-16 check by hand, and the register note and PR body match the code. Approving.
codecov/patch is failing on this head (other checks were still running when I looked), so it will need a look before merge.
Optional:
- Mixed inputs. A joint return where one spouse enters
capital_gainsdirectly and the other enters the short- and long-term parts adds back a loss AGI never saw (32,000 instead of 30,000). Montana's sibling formula guards that case. - Reforms. The capital and business add-backs still don't check that
loss_aldis deducted in AGI. - From the first round, still open: a dependent's tax-exempt interest,
#page=1on RCC-146, and a joint-return case.
Part of #10812 (state credit income audit).
What was wrong
vt_renter_credit_incomewas federal AGI + 75% oftax_exempt_social_security+ tax-exempt interest. 32 V.S.A. 6061(18) (statute) defines renter credit income as federal adjusted gross income "increased by" four items, and the model had only (B) and (C):tax_exempt_social_securitysums every tax unit member, so 75% of a dependent child's survivor benefits counted as the parent's income.What changes
limited_business_lossadded tovt_renter_credit_income: each head's and spouse's loss in each category, as federal AGI deducted itvt_renter_credit_capital_loss_add_backadds back every capital loss deducted in AGI: the net loss on Form 1040 line 7 up to the $3,000 limit (limited_capital_loss); losses absorbed by another filer's gains or by distributions without Schedule D (capital_losses_allowed_against_gains); and, for the head and spouse, losses absorbed by the same person's other-term gains, Schedule D line 13 distributions or against a prior-year long-term carryover. Read like the business losses, "capital loss" is each loss on a sale or exchange (26 U.S.C. 1222(2), (4)); RCC-146 line 13 uses the net line 7 figure (registered)gov.states.vt.tax.income.credits.renter.deduction_add_backsand variablevt_renter_credit_deduction_add_back. The list holds each (D) item the model has a variable for (educator expenses, HSA, deductible SE tax, SE retirement plans, SE health insurance, early withdrawal penalty, alimony paid, IRA, student loan interest, tuition and fees). Only items on this year's federal ALD list count ("to the extent excluded"), summed over the head and spouse asabove_the_line_deductionssums them, as Michigan's household resources dovt_renter_credit_countable_tax_exempt_ssreadsfiler_tax_exempt_social_security(head and spouse) instead oftax_exempt_social_securityThe 2021 RCC-146 has the same lines 10 to 14, and (18) took effect for 2021, so the change applies to every year the credit exists.
Source conflict (recorded as candidate
us-vt:renter-credit-loss-add-back-nettinginsource_conflicts/candidates.yaml). RCC-146 line 13 adds back a negative Schedule 1 line after the items on it net: line 5 combines rental, partnership, S corporation and estate items, and line 3 combines both spouses' Schedule Cs. The statute lists each kind of loss. Under the repo's statute-first rule the model adds back each head's and spouse's loss in each category, so a $4,000 partnership loss next to $10,000 of rental income is added back (Case 7 ofvt_renter_credit_income.yaml; line 13 would add nothing). The difference arises only when one return has income and a loss in categories that share a line. NoKNOWN_UNMARKEDline is needed: the variable's comment doesn't match the guard's patterns.#10647 (draft) lists the Vermont renter credit as a reader to review in step 3 of its plan; this PR settles it, and #10647's diff doesn't touch Vermont. Draft #10722 edits the same formula in
vt_renter_credit_countable_tax_exempt_ss.pyto add tier 1 railroad benefits; whichever merges second keeps both (filer_tax_exempt_social_securityplus the railroad term).Who is affected
Vermont renters who claim the renter credit (Form RCC-146). Only
vt_renter_creditmoves; no Vermont income tax variable reads it, and it reaches household benefits and SPM resources throughstate_property_tax_refunds. Income rises, and the credit falls or ends, for claimants with business, farm, rental, pass-through or capital losses, or with self-employment tax, IRA, HSA, student loan interest, alimony, educator or early withdrawal deductions. Every self-employed claimant with net earnings of $400 or more has the deductible half of self-employment tax. Income falls, and the credit rises, for a claimant whose dependent child or other dependent receives Social Security.Tests
Moved from #10812 (
vt_renter_credit_income_audit.yaml) to the newvt_renter_credit_income.yaml, values unchanged: Case 1 early withdrawal penalty (30,000), Case 2 business loss (30,000), Case 3 capital loss above the limit (30,000; the carryover is not added back), Case 4 dependent child's Social Security (20,000).Added, with arithmetic in each comment:
vt_renter_credit_income.yamlCases 5-10: capital loss exactly at the $3,000 limit (30,000); capital gain distributions offset by a capital loss (32,000: the loss they absorbed is added back too); partnership loss offset by rental income (40,000); deductible self-employment tax (20,000); IRA and student loan interest (30,000); a dependent's own early withdrawal penalty not added back (30,000).capital_gainsentered directly plus a carryover (32,000).vt_renter_credit_countable_tax_exempt_ss.yamlTest 3: a parent's 2,000 counts (1,500), the dependent child's 8,000 doesn't.vt_renter_credit.yaml, four end-to-end Addison 2025 households: a 2,000 business loss moves a claimant from the full 1,324 into the phase-out (1,250); with the add-back, income exactly at the 24,400 full credit limit (1,324) and 50 above it (1,322, AGI alone gave 1,324); income exactly at the 52,850 partial credit limit (0, AGI alone gave 23).Existing tests changed: inputs only, no expected value.
vt_renter_credit_countable_tax_exempt_ss.yamlTests 1 and 2 enteredtax_exempt_social_securitydirectly, which the formula no longer reads; they now enterfiler_tax_exempt_social_securitywith the same amounts (75 and 0).Not in this PR
tax_exempt_interest_incomeinaddssums every member, while line 12 reads the claimant's Form 1040 line 2a. The review doesn't list this; noted here only.axiom: needed (rulespec-us has no us-vt module for 32 V.S.A. 6061(18); no encoder run was authorized for this PR)
After the review fixes, tested locally:
vt_renter_credit_income.yaml(16), the Vermontcredits/rentdirectory (970) andtest_source_conflicts.py(251) pass. Before them, tested on Modal: runw1-vt-a, 239/239 passed (233 YAML files: the Vermont folder, every partner file, the changed files and the other files that readvt_renter_credit; plus 6 Python files:test_source_conflicts.py,test_reference_urls.py,test_parameter_yaml_keys.py, the two Vermont renter credit Python tests and the claimable-filer invariants). The follow-up commit only adds this PR's link to the candidate;test_source_conflicts.pypassed on it in runw1-vt-b(251 passed).🤖 Generated with Claude Code